Drug abuse is a chronic relapsing disorder. Key differences exist between drug reinstatement in laboratory animals and drug relapse in humans. The first objective of this proposal is to develop parallel laboratory models of relapse in non-treatment seeking human cocaine abusers and in nonhuman primates. We will approximate the clinical situation, where rates of drug taking start off high (baseline), then become low (treatment) and then often become high again (relapse), by using behavioral contingencies to produce low rates of cocaine intake. We will model the choices made by individuals in treatment to take drug by providing participants opportunities to choose between drug and an alternative reinforcer. Once we have established the behavioral contingencies needed to produce a low rate of drug taking, the second objective is to examine the effect of a single self-administered dose of cocaine on cocaine choice, i.e., cocaine- induced relapse. As in treatment, and in contrast to traditional reinstatement models, both humans and non- human primates will have access to cocaine, not placebo, during all test sessions. Aim 1: Determine a set of behavioral contingencies that will be effective in decreasing the choice to self- administer cocaine. We will increase the cost of choosing cocaine by increasing the work requirement to obtain a dose, and we will increase the value of the alternative reinforcer to cocaine by increasing the probability of earning money in a game of chance in humans, and a preferred food treat in nonhuman primates. We hypothesize that 1) cocaine choice will decrease with increasing cocaine cost and with increasing value of the alternative to cocaine use; and 2) a single dose of "free" cocaine (available without cost or alternatives) self-administered immediately before the choice session will increase the level of cocaine choice in a dose-dependent manner, i.e., cocaine-induced relapse. Aim 2: Provide "proof of concept" for the new model by determining the effect of maintenance on modafinil, an alerting agent that has demonstrated clinical utility in treating cocaine abuse, on cocaine choice and cocaine-induced relapse. We hypothesize that maintenance on modafinil will decrease cocaine choice and cocaine-induced relapse. We will model essential aspects of relapse: non-treatment seeking humans and nonhuman primate's will make decisions to initiate drug use based on how hard they will have to work for drug and how much they will have to forego in non-drug alternatives. The parallel design for humans and nonhuman primates will allow us to simultaneously refine current preclinical laboratory animal and human models of relapse, and thereby improve our ability to develop more effective strategies for decreasing relapse to drug use. [unreadable] [unreadable] [unreadable]